FAST THERMAL DESORPTION SPECTROSCOPY AND ULTRAFAST MICROCALORIMETRY: NEW TOOLS TO UNCOVER MYSTERIES OF ICE
Vlad Sadtchenko
Chemistry Department
The George Washington University
Washington, DC, 20052 USA
1. INTORODUCTION
Due to the crucial role of aqueous chemistry in a variety of environmental, biological, and industrial processes, experimental studies of ice remain an important field of modern physical chemistry. Examples of applied areas, which require knowledge of physics and chemistry of various solid forms of water, include: atmospheric chemistry and climate change, soil chemistry, planetary and interstellar chemistry, cryopreservation, and research into alternative energy sources. Furthermore, ice and water are uniquely accessible to computational modeling, due to the availability of extensive information on the intermolecular interactions in water-containing systems. Therefore, ice is often considered to be a model system for studies of fundamental properties of condensed molecular phases.
In spite of centuries of scientific inquire many fascinating properties of ice still await the comprehensive explanation. The lack of fundamental understanding of physics and chemistry of condensed aqueous phases is due to the numerous challenges involved in laboratory studies of such systems. Many of the standard analytical methods that have been developed to investigate reactions and dynamics in solids and liquids are not sufficiently sensitive, fast, or selective to provide detailed characterization of processes of interest in varying aqueous environments. The failure of contemporary experimental science to address many fundamental questions about nature of aqueous phases is reflected in the current jargon, which, for example, includes a notion of “No Man’s Land”, i.e., “experimentally inaccessible” temperature region in studies of properties of amorphous ice. Without a doubt, the development of new experimental techniques for studies dynamics and reactions in various condensed aqueous phases is long overdue, and, therefore, has been the focus of our research program at the George Washington University. Over the past five years, we have developed a set of novel experimental techniques, which we collectively refer to as Fast Thermal Desorption Spectroscopy - Ultrafast Scanning Calorimetry (FTDS-USC), and which we have already used to obtain data crucial for resolving several long-standing controversies in the field. The following sections of this article contain the description of our apparata, and the main venues of the research in our laboratory.
2. OVERVIEW OF EXPERIMENTAL TECHNIQUES
Over the past decade Thermal Desorption Spectroscopy (TDS), in its various modifications, has been widely used to study adsorption, desorption, diffusion, and reactions in thin (0.01-1 mm) aqueous films under high vacuum conditions 1-13. In these types of experiments, thin ice films of various phase composition, saturated with variety of chemical dopants are grown on a flat substrate, at cryogenic temperatures in vacuum using effusive or supersonic molecular beam sources. Because of low temperatures, at which the films are deposited, reactions usually do not proceed during film deposition. After the film growth, the temperature of the substrate is raised in a controlled fashion, and vaporization of the film (accompanied by diffusion and reactions of dopant species) is initiated. During film’s vaporization, a sensitive and selective analytical technique, or a combination of several such techniques are used to monitor chemical and physical phenomena in the film. For example, in Temperature-programmed Reaction Mass Spectrometry studies, the volatile chemical species, which evolve from the film, are monitored with a sensitive mass spectrometer.
There are several factors, which make TDS-based approach particularly suitable for experimental studies designed to gain molecular-lever information on ice properties. First of all, a TDS approach allows precise control of film’s morphology and phase composition. Indeed, an amorphous microporous, amorphous solid, crystalline cubic, or crystalline hexagonal films can be grown on the variety of substrates1-12. Second, deposition of ice at cryogenic temperatures makes it possible to create intricate, non-equilibrium, spatial and chemical distributions of dopant species in the film. For example, a sandwich-like film consisting of various isotopes of water can be easily grown. The isotopic scrambling in such a film at higher temperatures has been used to determine transport properties of ice on nanoscale6,8. Third, a wide variety of analytical techniques can be applied in order to monitor chemical transformations in thin ice films. Examples of analytical techniques used to study aqueous surface at cryogenic temperatures are numerous and include mass-spectrometry, vibrational spectroscopes, photoelectron spectroscopy, atomic force microscopy, proton scattering etc. Finally, due to high vacuum conditions characteristic of TDS experiments, surface and condense phase phenomena are not obscured by processes in the gas phase.
In spite of all these powerful features, the TDS approach has its obvious limitations. In the case of ice, chemical phenomena can be studied with TDS technique only at temperatures below -80 0C, where volatility of this molecular solid is relatively low. One of the particularly nasty practical problems arising during attempts to extend the temperature range of the TDS studies is the formation of gas boundary layer above ice free surface. Such a layer forms as a result of rapid vaporization of macroscopic ice sample and is due to collisions and back scattering of water and other desorbing molecules. Thus, the formation of boundary vapor layer above the surface of ice samples in high temperature TDS experiments eliminates the main advantage of this approach, i.e., its ability to exclude the gas phase phenomena from interpretation of the experimental results. Aqueous phase phenomena at temperatures below -80 0C are, indeed, of great fundamental as well as of applied significance. Nevertheless, due to likelihood of fundamental changes in the nature of aqueous interfaces near ice melting point14-16, the results of low temperature studies may not be automatically extrapolated to higher temperatures typical for many natural environments.
Several years ago, we found a solution to the problems listed above. The central idea of our experimental approach was inspired by the pioneering work of Faubel and Kisters17 who reported TOF spectra of H2O and CH3COOH molecules evaporating at near ambient temperature from surfaces of 10-50 mm thick jets of water injected into a vacuum chamber. These TOF measurements showed that when the diameter of the water jet was reduced from 50 to 10 mm, the TOF distributions changed from that characteristic of a supersonic molecular beam to a Maxwellian distribution representative of the collision-free molecular flow. Using this important result, we have achieved a critical improvement of TDS technique by changing the substrate geometry18,19. Instead of using a flat substrate for ice film deposition, we grow ice on a thin (~ 10 mm) tungsten filament. A number of advantages instantly follow:
First, because desorbing molecules undergo only a few gas phase collisions formation of a dense boundary vapor layer over the ice sample surface does not occur at temperatures up to 0 0C. [21].
Second, the small surface area of ice films grown on a thin wire results in a small net flux of desorbing molecules making it possible to keep the load on vacuum system at manageable levels. [21]. Third, due to small mass of thin wire, heating rates in excess of 105 0C·s-1 can be easily achieved making it possible to bring temperature of a few micrometer thick aqueous films to a value as high as 0 0C before vaporization of any significant fraction the ice sample [21]. Finally, because the mass of the microscopic filament is small, and the heat capacity of tungsten is about an order of magnitude lower than the specific heat capacity of any condensed aqueous phase, scanning calorimetry can be used to measure the variations in the heat capacity of the aqueous film during rapid heating. As we discuss later in this article, high heating rates achieved in our experiments provide a unique opportunity for studies of molecular kinetics in amorphous ice [21].
A detailed description of apparatus and standard experimental procedures can be found in our publications [21]; therefore, only a brief account is given here. As we have already mentioned, ice films are grown on the surface of a tungsten filament (10 mm in diameter, 2 cm long), which is spot-welded to the supports of the filament assembly. The supports, while in thermal contact with the liquid nitrogen-cooled heat sink, are electrically isolated from the rest of the apparatus. The filament assembly includes a thermal control system capable of maintaining the filament supports at temperatures in a range from -160 0C to -120 0C. The assembly is surrounded by cryogenic shields and positioned in a vacuum chamber pumped with a 2500 s-1 Varian diffusion pump. The cryogenic shields significantly reduce the gas load on the pumping system of the apparatus, allowing it to maintain pressures below 3∙10-7 Torr during all stages of FTDS experiments.
Water vapor and dopants are delivered from the vapor source to the filament via 12 effusive dosers made of 1/8’’ stainless steel tubes, which are equally spaced around the filament at a distance of approximately 5 cm from its center. The intersecting vapor beams from the dosers leads to the formation of a relatively dense H2O vapor cloud around the tungsten filament, and, thus, facilitates the deposition of uniform ice films with a neat cylindrical geometry. After deposition, rapid isothermal vaporization of the film is initiated by applying a 2V – 5V potential difference across the filament. In the course of the entire FTDS experiment, the voltage across the filament and the current through the filament are recorded every four microseconds by a custom designed data acquisition system. The temperature of the filament is then calculated from the resistance data using the temperature coefficient of electrical resistivity of tungsten. During FTDS experiments, the kinetics of the vaporization of various chemical species are monitored simultaneously with three detectors: a quadrupole mass spectrometer (QMS) positioned 10 cm away from the filament, a QMS positioned 95 cm away from the filament, and a Fast Ionization Gauge (FIG), positioned 6 cm away from the filament. The apparatus also employs a single-slit chopper disc, which facilitates Time-of-Flight characterization of the velocity distribution of the vaporization products.
Figures 1 illustrates the basic thermal desorption experiment [21,22]. It shows the H2O flux from the filament and the corresponding temperature of the filament, as determined from resistance data. The H2O flux was measured with the fast ionization gauge. As shown in the figure, applying a small (3.5 V) electrical potential across the filament result in rapid heating of the ice-filament system during the first millisecond of the experiment. Nevertheless, after approximately 2 ms, the filament achieves the steady-state vaporization regime which is characterized by nearly constant temperature and H2O vaporization flux. At this stage of the experiment, the heat generated by the filament is carried away by vaporization of ice from its surface. The film vaporization follows zero-order kinetics for the next 5 ms, i.e., until 80% of the ice sample is gone, then the film breaks into clusters and islands on the filament surface resulting in non-zero kinetics, and the increase in the filament temperature.
 |
Observation of zero-order vaporization kinetics leads to several important conclusions. First, during steady-state vaorisation near 00C, the ice adlayer on the surface of the filament is a neat film free of large pores or ruptures, i.e., the film covers the entire filament surface [21,22]. Second, the temperature gradient along the length of the filament is negligible except within a fraction of a millimeter near the ends of the filament [21]. Finally, since the temperature of the filament does not vary significantly with the film thickness during vaporization the temperature gradient in the film must be less than a few degrees [21,22]. Because the typical time scale of our desorption experiments is up to four orders of magnitude shorter that that in classical TDS studies, we have termed our technique, Fast Thermal Desorption Spectroscopy or FTDS.
 |
Finally, we would like to explain the basics of our Ultrafast Scanning Microcalorimetry technique, which is an integral part of the FTDS apparatus, and which makes it possible to determine mass and phase composition of ice samples [21,22]. In USC experiments, the combined effective heat capacity of the ice film and the filament is calculated as the ratio of the power generated by the filament to the first time derivative of the temperature. Figure 2 illustrates typical MSC data. The dashed line shows the effective heat capacity of the ice-free filament; the solid line shows the effective heat capacity of the filament with an H2O film vapor-deposited at – 120 oC; and the dotted line shows the effective heat capacity of the filament with an H2O film vapor-deposited at -150 oC. In the case of deposition at -120 oC, the monotonic increase in heat capacity of the filament-film system, i.e. the lack of exothermic transitions characteristic of crystallization, demonstrates that the ice sample is crystalline [21]. However, in the case of films deposited at -150 oC, the effective heat capacity of the film shows partially overlapping irreversible exotherms, which are due to crystallization of the initially amorphous ice samples [21]. Assignment of the exotherms shown in Fig. 4 to crystallization of amorphous ice is facilitated by the 2 kJ/mol enthalpy release, which is close to previously reported values [20]. These calorimetric experiments also show that non-crystalline ice films always undergo crystallization before steady state vaporization can occur at temperatures above -25 oC. In addition to providing information on the phase composition of ice samples, the data shown in Fig. 4 were used to determine the masses of the ice films. Assuming that the film has the specific heat capacity and density similar to those of hexagonal ice we can estimate the thickness of ice films with accuracy of ± 10 % [21,22].
3. CURRENT RESULTS
In addition to the well-known benefits of thermal desorption spectroscopy, the apparata described in the previous section has another important advantage. Unlike any other technique, the FTDS-USC setup, in its full configuration, makes it possible to study ice in the entire temperature range from cryogenic to near ambient. While information on phase transitions, transport phenomena, and reactions near 0 0C can be studied with FTDS directly, USC can be employed to investigate phenomena in amorphous ice at cryogenic temperatures. Here, we provide review of the current research projects in our laboratory. We begin with USC investigations of molecular kinetics in amorphous ice.
Ultrafast microcalorimetry studies of low density amorphous ice. Water is the most important yet unusual liquid found in nature. Its structure and properties are not well understood in spite of centuries of scientific inquiry. The anomalous properties of water are most noticeably manifested when water is supercooled below its equilibrium freezing point. The constant-pressure, specific heat capacity, the coefficient of thermal expansion, and the isothermal compressibility of water begin to diverge as the temperature approaches 228 K (Ts). Development of realistic models of the structural transformations in water upon supercooling require measurements of its properties (i.e. diffusivity, viscosity, heat capacity, etc.) at temperatures near and below Ts. However, such measurements have proven to be a great experimental challenge. Unless the cooling rate is exceedingly high (107 K / second) and the sample dimensions are microscopic, crystallization becomes inevitable, as the temperature of supercooled water is lowered to the vicinity of Ts.
An alternative approach to the measurements described above utilizes glassy water samples grown by slow H2O vapor deposition. The low density amorphous (LDA) ice samples, grown at cryogenic temperatures and heated above the postulated glass transition, offer a way to obtain measurements of supercooled water properties at temperatures below Ts. Unfortunately, as the temperature of an LDA sample is raised to about 150-160 K, the LDA ice rapidly crystallizes to form cubic ice. Thus, the characteristic temperature of rapid crystallization of low density amorphous ice (Tc ~ 160 K) and the temperature of catastrophic crystallization of supercooled water (Ts ~ 228 K) mark the borders of a temperature region dubbed “No Man’s Land”, where experimental studies of non-crystalline states of water have been deemed to be impossible [2].
With the objective of gaining insights into the properties of deeply supercooled water, we utilized the capabilities of our ultrafast microcalorimetry technique to investigate thermal dynamics in LDA ice, under conditions of ultrafast heating. Due to high heating rates used in our experiments, we were able to conduct measurements of LDA heat capacity at temperatures as high as 205 K. The thermogram in Fig. # shows the most important results of our experiments. It presents the measured heat capacities of LDA ice at temperatures significantly higher than Tc. The heat capacity of a crystalline (cubic) ice sample of the same mass is shown for comparison (dotted line). The exothermic peaks in the thermogram of the LDA sample are due to its crystallization and prove that our LDA samples contain a low fraction of crystalline ice. The endothermic upswing in heat capacity is due to the onset of rapid ice vaporization at temperatures above 220 K. The striking feature of the thermogram shown in Fig. 3 is the lack of significant differences in the heat capacities of amorphous and crystalline aqueous phases at temperatures between 160 and 205 K, i.e. above the glass transition temperature of 136 K clamed in some previous studies [3-5]. According to the predictions of heat capacity dependence on temperature, which are based on the glass transition temperature of 136 K, the heat capacity of LDA must rapidly deviate from that of crystalline ice. Our measurements show less than a few percent difference between heat capacities of LDA and cubic crystalline ice at temperatures up to
 |
205 K, which is in direct contrast with the predictions.
In order to demonstrate that it is actually possible to observe thermal phenomena characteristic of supercooled liquids with our microcalorimetry approach, we have conducted measurements of the heat capacity of LDA samples contaminated with acetic acid. Figure 3 compares the thermogram of pure LDA ice to that of LDA/Acetic Acid mixture (10:1). While the heat capacity of the LDA/Acetic acid mixture undergoes a rapid increase in heat capacity at 175 K, the heat capacity of pure LDA remains nearly equal to that of crystalline ice.
 |
The data shown in Fig. 4 makes it possible to estimate the enthalpy relaxation time in the case of the LDA/Acetic acid mixture at 175 K (i.e. at the onset temperature of heat capacity rise). Taking into account the value of the heating rate in our experiments (105 K/s), we arrive at the conclusion that the relaxation time is on the order 10-5±0.5 s at 175 K for the LDA/Acetic acid mixture. Because we do not observe any noticeable upswing in the heat capacity of pure LDA sample, we conclude that even at temperatures as high as 205 K the enthalpy relaxation time for pure LDA must be below 10-5±0.5 s, which is three orders of magnitude greater than that obtained assuming the existence of glass transition for pure LDA at 136 K [6]. In summary, our ultrafast microcalorimetry experiments provide evidence to support of the recent arguments by Angel et. al. [7] that the glass transition in LDA must occur at temperatures much higher than 136 K. Furthermore, our experimental results emphasize the potential of contaminants to lower the glass transition temperature of LDA.
FTDS Studies of vaporization kinetics of aqueous films. In 1971 Davy and Somorjai presented their measurements of the desorption rate of ice up to – 40 0C [43]. These experiments have laid a foundation to the precursor mediated mechanism for ice desorption. The model predicts zero-order desorption kinetics in two limiting cases. At low temperatures, the mobile precursor’s (MPs) desorption rate from the surface is low and the overall ice vaporization process is governed by pre-equilibrium between the MPs and the surface molecules. The effective activation energy is approximately the desorption enthalpy, which is approximately 50 kJ/mole. At higher temperatures, the rate of desorption of the MPs increases drastically destroying the equilibrium between the MPs and surface molecules. The formation of the mobile desorption precursors becomes the limiting step of the desorption process, which results in the effective activation energy of half the desorption enthalpy. Although many highly sophisticated measurements of ice vaporization rate were attempted in the past none of them were conducted at temperatures above - 40 0C.
Figure 5 shows the first direct measurements of absolute vaporization rate of crystalline D2O and H2O ice at temperatures above -40 0C obtained in our FTDS experiments [21]. Different symbols represent results from experiments with ice films of distinct thermal history and thickness. The solid lines in the Fig. 5 show vaporization rate values calculated from ice equilibrium vapor pressure under the assumption that the mass accommodation coefficient is equal to unity, i.e. that the vaporization rate is equal to the maximum equilibrium rate given. The dotted lines show the range of possible desorption rate values predicted by the simple MP mechanism.
Data shown in Fig. 5 lead to the following conclusions: at temperatures near 0 0C the desorption rate demonstrates Arrhenius behavior with the effective activation energy of 50 ± 4 kJ/mol, which significantly exceeds the value predicted by the simple MP mechanism (25-35 kJ/mol). Third, at temperatures above -40 0C, the ice vaporization rate greatly exceeds the desorption rate predicted by the simple MP model. In summary, the mobile precursor mechanism, as formulated by Somorjai and Davy, fails to describe the desorption kinetics of ice at temperatures near its melting point.
In the past we argues that the deviation of observed ice vaporization kinetics from a classical MP model can be explained by surface roughening or premelting transition at temperatures above - 40 0C. AlRecent experimental studies have emphasized the complexity of water/vapor interface in respect to the dynamics of gas uptake and release which is now treated in terms of the nucleation theory [45]. Our work indicates that uptake and release of gases on ice/vapor interface
 |
at temperatures near ice melting point may be equally or even more complex as that on liquid surfaces. Because the uptake and release of gases on ice is likely to depend on ice surface
morphology, theoretical description of these processes may require a new formalism that takes into account possible surface phase transitions at various temperatures. Many questions, however, remain unanswered. For example, it is still unclear at what temperature the onset of the premelting actually occurs. Is the formation of liquid-like layer preceded by a roughening transition? How does it affect the ability of ice to trap and release various gases? Is the interface between vapor and quasiliquid layer different from neat water surface? These questions mark an exciting area for the future studies and emphasize the importance of development new experimental methods capable of probing surface dynamics of volatile molecular solids near their melting points. Using our Fast Thermal Desorption technique we will continue our investigation into vaporization kinetics of condensed aqueous phases. The next step in our work will be extensive investigations of vaporization kinetics of ice films doped with a variety of environmentally relevant impurities.
Studies of transport phenomena, interfacial phase transitions, and reactions in polycrystalline ice near its melting point. In order to demonstrate that Fast Thermal Desorption spectroscopy can be used successfully to study reactions in volatile polycrystalline materials, and to gain insights into nanoscale molecular transport in polycrystalline ice, we have conducted preliminary studies of H/D exchange kinetics near ice melting point. Figure 10 illustrate our approach. As shown in the Figure 10, a thin layer of D2O, initially positioned at a distance away from the film surface, evolves from the ice film only after vaporization of the overlaying H2O16. The time interval from the onset of H2O vaporization to the appearance of the D2O vaporization peak, i.e., the mean residence time of D2O in the bulk of the H2O film, gives the mean reaction time for H2O/D2O isotopic exchange. Positioning the D2O layer further away from the surface, i.e. increasing the mean reaction time, results in a gradual decrease of the magnitude of the D2O peak. The decrease in the D2O peak is accompanied by a gradual increase in the HDO yield signifying rapid conversion of D2O to HDO.
 |
Figure 6 shows the H/D exchange kinetics between 50 nm thick D2O layer and the surrounding H2O16 polycrystalline ice at -5 0C. As shown in the figure, only half of the initial D2O layer is converted into HDO on the time scale of our FTDS experiment. Analysis of the isothermal desorption spectra of HDO along with the observed reaction kinetics for D2O layers of various thickness show that H/D exchange is controlled by the rate of inter-diffusion of H2O and D2O in polycrystalline ice. According to our results, the H/D exchange reaction is limited to relatively narrow zone at H2O/D2O interface. Because the width of the reaction zones must be defined by the inter-diffusivity of H2O and D2O and because the HDO and D2O yields depend on the dimensions of the reaction zone, we can estimate the characteristic scale of inter-diffusion of D2O and H2O on the time scale of our experiments. According to our estimates, the characteristic diffusion scale over 3 ms diffusion/reaction time is less that 50 nm at -2 0C. In the future, we will conduct similar experiments in the entire temperature range accessible with our Fast Thermal Desorption technique. Furthermore, we plane to investigate influence of various environmentally relevant impurities on the water self-diffusivity in polycrystalline ice at various temperatures. We are confident that, combined with a appropriate theoretical treatment, such data will make it possible to gain valuable insight into grain boundary premelting [46] and other morphological dynamics in important condensed phase system.
 |
In summary, we demonstrate that Fast Thermal Desorption Spectroscopy can be used to study kinetics of reactions in polycrystalline ice near its melting point, obtained detailed information on nanoscale molecular transport in polycrystalline ice at temperatures near its melting point.
4. AFTERWORD
Several experiments reviewed in this article illustrate just a small fraction of opportunities for research into fundamental physical and chemical properties of ice offered by our experimental approach. We emphasize that our core method, i.e., the FTDS can be combined with a variety analytical techniques. For instance, the existing apparatus can be equipped with an FTIR spectrometer for better initial characterization of phase and chemical composition of ice films during deposition. At the present time we are also developing an optical system that would allow us to conduct detailed studies of photochemical reactions in polycrystalline ice. While our USC measurements were conducted a single heating rate, further improvement of this particular component of our apparatus will make it possible to measure enthalpy relaxation times in doped polycrystalline ice for a wide range of temperatures. Finally, we are also in the process of designing a novel data acquisition system which would facilitate FTDS and USC measurements with ice film subjected to strong electrostatic fields (in excess of 109 V/m). We hope that at some point the FTDS-USC approach will become a standard tool for studies of reactions and dynamics in ice and other volatile molecular solids.
References
1. F. E. Livingston and S. M. George, J. Phys. Chem. 1998, A 102, 10280.
2. R. S. Smith, C. Huang, E. K. L. Wong, and B. D. Kay, Phys. Rev. Lett. 1997, 79, 909.
3. J. Gordon Davy and G. A. Somorjai, J. Chem. Phys. 1971, 55, 3624.
4. D. R. Haynes, N. J. Tro, and S. M. George, J. Phys. Chem. 1992, 96, 8502.
5. N. J. Sack and R. A. Baragiola, Phys. Rev. 1993, B 48, 9973.
6. D. E. Brown, S. M. George, C. Huang, E. K. L. Wong, K. B. Rider, R. S. Smith, and B. D. Kay, J. Phys. Chem. 1996, 100, 4988.
7. R. J. Speedy, P. G. Debenedetti, R. S. Smith, C. Huang, and B. D. Kay, J. Chem. Phys. 1996, 105, 240.
8. J. A. Smith, F. E. Livingston, and S. M. George, J. Phys. Chem B. 2003, 107, 3871.
9. V. Sadtchenko, W. R. Gentry, and C. Giese, J. Phys. Chem. B, 2000, 104, 9421 ().
10. J. D. Graham and J. T. Roberts, J. Phys. Chem. 1994, 98, 5974.
11. A. Bar-Nun, I. Kleinfeld, and E. Kochavi, Phys. Rev. B 1988, 38, 7749.
12. S. Haq, J. Harnett, and A. Hodgson, J. Phys. Chem. B, 2002, 106, 3950.
13. Reference 1-12 represent modern examples of TDS approach and not an extensive review of the field.
14. J. G. Dash, in Ice Physics and the Natural Environment, edited by J. S. Wettlaufer, J. G. Dash, and N. Untersteiner, NATO ASI Series I Vol. 56 (Springer-New York, 1999), p. 11.
15. J. S. Wettlaufer, Philos. Trans. R. Soc. London, Ser. A 1999 357, 1.
16. J. S. Wettlaufer, Dynamics of ice surfaces, Interface Science, 2001, 9, 115.
18. V. Sadtchenko , M. Brindza, M. Chonde, B. Palmor, R. Eom, J. Chem. Phys., 2004, 121 (23), 11980., and references therein.
19. Haiping Lu, Stephanie A. McCartney, M. Chonde, D. Smyla, and Vlad Sadtchenko*, J. Chem. Phys., 2006, IN PRESS